Skip to contents

Generate a variable correlated with an existing one

correlatedValue.Rd

Draws a new variable y = r*x + noise, with noise scaled so that cor(x, y) is approximately r. Negative results are clipped to 0 (this assumes x/y are non-negative quantities, e.g. pigment concentrations – see get_distributionLUT's DepCab option, which uses this to correlate Car with Cab).

Usage

correlatedValue(x, r)

Arguments

x

numeric vector. The variable to correlate against.

r

numeric (-1 to 1). Target correlation coefficient between x and the result.

Value

A numeric vector the same length as x, correlated with it at approximately r.

Examples

Cab <- runif(200, 10, 80)
Car <- correlatedValue(x = Cab / 4, r = 0.8)
cor(Cab, Car)  # close to 0.8
#> [1] 0.9883911